Information about Asia and the Pacific Asia y el Pacífico
Journal Issue
Share
Article

Philippines: Selected Issues

Author(s):
International Monetary Fund. Asia and Pacific Dept
Published Date:
August 2014
Share
  • ShareShare
Information about Asia and the Pacific Asia y el Pacífico
Show Summary Details

Philippine Inflation: Home Grown or Imported?1

A. Introduction

1. With the exception of the global financial crisis (GFC), world inflation has generally remained low since 2000. In addition, cross-sectional variation in inflation across different regions and types of economies has declined noticeably in recent years (Figure 1). In the Philippines, inflation has also moderated despite strong GDP growth and high capacity utilization rates, and has been closely tracking world inflation. To what extent does the Philippines’ inflation performance reflect domestic macroeconomic conditions and policies, or is it mainly determined by global developments? This chapter addresses this question by applying a latent factor model to the inflation rates of 62 countries to identify the global common drivers of inflation and their importance for individual countries. Taking into account these findings, the paper also adopts a more standard single-country, single-equation model to assess the drivers of Philippines’ inflation.

Figure 1.Total Inflation, 2000–13

(In percent, seasonally adjusted, period average, year-on-year change)

Source: IMF, World Economic Outlook.

2. We find that global common factors account for about 60 percent of the variation in Philippine inflation (about the average for the country sample), with country-specific effects explaining the remainder. From the single equation model we find that exchange rate pass-through, world commodity prices and the output gap, together with lagged own inflation explain most of the Philippines’ inflation.

3. The paper is organized as follows. The following section discusses inflation developments and the monetary policy framework in the Philippines. Section 3 employs a global latent factor model to decompose inflation into common drivers and idiosyncratic factors for a sample of 62 countries. Based on these results, it then models inflation in different regions. Section 4 presents the single-country, single equation model and conducts out of sample forecasts to determine consistency with the medium-term inflation target.

B. Philippines: Inflation Developments and the Monetary Policy Framework

4. Since the adoption of an inflation targeting framework in January 2002, Philippines’ inflation has averaged 4.3 percent, considerably lower than the 9.2 percent during the 1990s. Nonetheless, for much of the period, inflation was outside the target band, although deviations have narrowed since the GFC, partly reflecting modest periodic adjustments to the inflation target bands. During 2012–13, inflation tended to lie near the bottom of the target band (4±1 percent) (Figure 2).

Figure 2.Philippines: Total Inflation and Inflation Target Bounds, 2001:M1-2014:M4

(In percent, period average, seasonally adjusted, year-on-year change)

Sources: Bangko Sentral ng Filipino; and Haver Analytics.

5. However, headline inflation picked up to 4.1 percent in December and reached 4.5 percent in May 2014, on typhoon-related food supply disruptions, reversal of the peso from previous appreciation to depreciation, higher administered rice prices, and accommodative monetary conditions. Among the Philippines’ near-neighbors, inflation increased very strongly in Indonesia, by somewhat less in Malaysia, but moderated in Thailand (Figure 3).

Figure 3.Total Inflation: Philippines and Its Major ASEAN Neighbors. 2000:Q1-2013:Q4

(In percent, seasonally adjusted, period average, year-on-year change)

Source: IMF, World Economic Outlook

6. From January 2015, the BSP will lower the midpoint of the target band to 3 percent, while keeping constant the width at ±1 percent. This will align the Philippines’ inflation target band closer to those of other advanced emerging markets (Figure 4), helping to limit real appreciation due to persistent inflation differentials. It is therefore important to determine the drivers of inflation and whether the new lower target can be achieved without compromising GDP growth.

Figure 4.Countries Operating a Fully Fledged Inflation Targeting Regime and the Current Target Bounds

(In percent)

Sources: Bank of England Handbook No.29, dill Hammond (2012); and Bank of Thailand.

7. Since January 2002, the Philippines has conducted monetary policy under an inflation targeting framework. This is consistent with the primary objective of the Bangko Sentral ng Pilipinas (BSP) “to promote price stability conducive to a balanced and sustainable growth of the economy.” The seven-member Monetary Policy Board is the decision-making body that sets key policy interest rates and other monetary policy instruments (including reserve requirements). The BSP immediately announces its policy decisions, publishes the minutes of the meetings with a four-week lag, and prepares a quarterly inflation report.

C. Common-Factors Analysis of Global Inflation

8. From Figures 1 and 3, it is apparent that inflation across different countries and regions has become more synchronized, especially since the GFC. A common factors model (CFM) can be used to identify the unobserved common factors that explain the observed comovement of inflation. In essence, factor models transform a large number of covarying series into a smaller number of orthogonal series (sequences of common factors) in such a way that each successive factor explains as much as possible of the remaining variation in the observed series. These observed series can therefore be expressed as the weighted sum of the common factors (the common-origin component), plus an idiosyncratic disturbance term, which is uncorrelated with the common-origin component and, hence, is country specific.

9. Common factor (also referred to principal component) models are widely used when the number of series is large relative to the time dimension of the data, a situation that standard estimation techniques cannot handle.2 This reflects that CFMs account for the variation in the large number of observed series by identifying a small number of common factors. For example, Sargent and Sims (1977) found that just two dynamic factors could explain a large fraction of the variance of U.S. quarterly macroeconomic variables, including output, employment, and prices. Similarly, Stock and Watson (1989) concluded that one factor was sufficient to model the comovements of major macroeconomic aggregates.

10. More recently, CFMs have been employed to analyze the behavior of international asset prices and inflation. For example, Choueiri and others (2008) decompose inflation in the EU-25 countries into common-origin and idiosyncratic components, with the aim of exploring the determinants of cross country differences in the role of common-origin inflation. Ciccarelli and Mojon (2010) investigate CPI inflation rates in 22 OECD countries using a similar strategy.

Data and Methodology of Common Factor Inflation Model

11. Our dataset covers quarterly inflation in 62 countries (see Appendix 1) and spans the period 2000:Q1 to 2013:Q3. Inflation is measured as the year-on-year percent change in the quarterly average of headline CPI. Data are seasonally adjusted and standardized by subtracting their means and dividing by their standard deviations.

12. The econometric methodology we employ entails five steps: (i) factor representation of the data; (ii) identification and estimation of the common factors: (iii) determination of the number of common factors; (iv) estimation of loading vectors that indicate the individual country’s responsiveness to each common factor; and (v) variance decomposition for each country.

Factor Representation

13. The cross country inflation can be represented in factor form as follows:

The expressions in (1) show that for each country i demeaned total inflation, πi,tπ¯i,t, can be decomposed into a common component (βft) and an idiosyncratic component (et), where f is a (K x 1) vector of latent (unobserved) factors, β is an (N × K) matrix, representing the loading coefficients, or weight of each common factor in each country’s inflation.

Estimation and Selection of Common Factors

14. Following Bai and Ng (2013) and Stock and Watson (2002), we estimate the static common factors in the panel by the method of asymptotic principal components. In addition to standard assumptions, identification of the common factors relies on the assumption that the factors are orthogonal to the idiosyncratic terms:

Extraction of the factors and their loadings is achieved by:

15. This methodology extracts multiple, ordered common factors, with each successive factor explaining less of the aggregate variance.3 The first common factor is found to explain 34 percent of the total variability of total inflation across the full sample of countries, with the second and third common factors explaining 18 percent and 10 percent, respectively. An information criterion that utilizes this property of diminishing explanatory power is used to identify the appropriate number of common factors that adequately describe the common-origin variance in the data. Intuitively, there is a trade-off between the benefit of including an additional factor in terms of increasing explanatory power and the cost of increased sampling error from estimating an additional parameter. Using the Bai and Ng (2002) criterion, the first three common factors—which collectively explain 62 percent of the variability of headline inflation—is found to be appropriate. Figure 5 shows the first three common factors extracted from the global inflation dataset.

Figure 5.Estimation of Three Latent Common Factors, 2001:Q1-2013:Q3

(In percent year-on-year change)

Sources: IMF, World Economic Outlook; and IMF staff calculations.

Factor Identification: What are the Common Factors for Global Inflation?

16. As noted, these common factors are unobservable statistical constructs. Therefore, to increase the usefulness of this approach, we investigate whether they can be associated with actual economic variables that theory suggests might influence global inflation:

  • Common factor 1 appears to fit well the behavior of global aggregate commodity prices, measured in U.S. dollars (Figure 6). The “fit” is especially strong since 2006, corresponding to a period with large swings in the prices of food and fuel.

  • Common factor-2 displays a downward movement during 2001–07, with stabilization thereafter. This pattern mirrors the decline in inflation that occurred across emerging market economies as they continued to stabilize following their transition from central planning to free markets or recover from widespread emerging market crises in the previous decade (Figure 7). This factor is therefore consistent with the “great moderation” in inflation.

  • Common factor-3 can be associated with developments in the nominal effective exchange rate of the U.S. dollar (Figure 8). This is consistent with the fact that the U.S. dollar is the numeraire for much of international trade, and that shifts in its NEER are passed through to local prices.

Figure 6.Common Factor-1 Analysis: The Possible Explanatory Variables, 2001:0.1-2013:0.3 1/

(In percent, year-on-year change)

Sourse: Imf, World Economic Outlook; and IMF Staff calculations.

1/ Data are standardized by subtracting the sample mean and dividing by sample standared deviation to ensure that high-variance series do not dominate results.

2/ All commodity price index, 2005=100, includes fuel and nonfuel prices indices; nonfuel price index, 2005=100, includes food and beverages and industrial inputs price index; fuel (energy0, 2005=100, includes crude oil(petroleum), natural gas, and coal price indices.

Figure 7.Common Factor-2 Analysis: The Possible Explanatory Variables, 2001:0.1-2013:0.3

(In percent, year-on-year change)

Source:IMF,World Economic Outlook, and IMF staff calculations.

Figure 8.Common Factor-3 Analysis: The Possible Explanatory Variables, 2001:0.1-2013:0.3

(In percent, year-on-year change)

Sources: IMF, World Economic Outlook; and IMF staff calculations.

Loading Vectors

17. The importance of a specific common factor for inflation can vary by country. This sensitivity is given by the loading vectors or weights which, in this model, are assumed to be static (i.e., constant over the entire period) and loaded contemporaneously. A higher loading indicates that the country’s inflation is influenced more heavily by that common factor. Loadings may also be negative. Differences in inflation across countries may therefore reflect not only country-specific factors, but differences in the loadings of common factors. Loadings may differ across countries for various reasons: the weight of food and fuel—both direct and indirect—in the consumption basket varies considerably across countries. In addition, responsiveness of domestic prices to world prices depends on the degree of competition in the domestic market, the existence of price-smoothing policies (e.g., subsidies or buffer stocks), the rigidities in labor markets that can amplify commodity price shocks, and the willingness of policymakers to dampen the price effects and the corresponding effectiveness of such policies.

  • Figure 9 reports the loading for each country for the first common factor. For nearly all countries, the loading is positive—as would be expected.

  • For the Philippines, with its relatively high weight of fuel and food in the CPI basket (Figure 10), the loading is a relatively high 0.667.

  • For the second common factor (“great moderation”), there is considerably more variation in loadings across countries (Figure 11). The loadings tend to be large for CIS, CEE and Latin American countries, capturing the substantial disinflation these countries achieved during the 2000s. Meanwhile, loadings tend to be small for most advanced economies. For Asia, negative loadings may reflect cost-push price pressures within regional supply chains. Consistent with other Asian countries, the loading for the Philippines is negative, though fairly small.

  • For the third common factor (U.S. dollar nominal effective exchange rate), loadings reflect the net effect of how the local currency reacts to a change in the U.S. dollar NEER and the pass-through to inflation of the change in the local currency (Figure 12). For example, loadings for emerging market countries tend to be positive, suggesting that their currencies weaken in response to U.S. dollar appreciation, which is then fed through to domestic prices.4 The loading for the Philippines is positive and relatively large.

Figure 9.Loading Coefficients of Factor-1

Sources: IMF, World Ecorromic Outlook and IMF staff calculations.

Figure 10.Philippines: CPI Basket, 2013

(In percent)

Source: Philippines, National Statistics Office

Figure 11.Loading Coefficients of Factor-2

Sources: IMF, World Ecorromic Outlook and IMF staff calculations.

Figure 12.Loading Coefficients of Factor-3

Sources: IMF, World Ecorromic Outlook and IMF staff calculations.

Variance Decomposition

18. As noted earlier, the first three common factors explain 62 percent of the total variability in global inflation. However, the explanatory ability of the common factors will vary across individual countries, depending on the importance of country specific shocks (Figure 13). At the upper end, common shocks account for around 90 percent of headline inflation variability, while at the low end, common factors explain little more than 10 percent of inflation variability. For the Philippines, common components account for 60.7 percent of headline inflation, with the remaining 39.3 percent therefore attributable to idiosyncratic disturbances.

Figure 13.Variability of Total Inflation Explained by the Common Factors

(In percent)

Sources: IMF, World Economic Outlook:, and IMF staff estimates.

19. Combining the common factors and their country-specific loadings, one can construct the level of common-origin inflation and derive idiosyncratic inflation for each country. This is shown in Figures 1416 for the Philippines, and in Figures 1720 for the ASEAN-5 countries. Common factors capture very well inflation developments in the Philippines, Malaysia, Thailand and Singapore. However, in Indonesia, common factors only explain 36 percent of inflation variability, against 61–73 percent for the other ASEAN-5. This reflects low explanatory capacity of common factors for Indonesian inflation during two episodes—late 2005-early 2006 and since mid 2013—corresponding to periods when energy price subsidies were significantly lowered. Nonetheless, common-origin inflations are very similar across the ASEAN-5 (Figure 21).

Figure 14.Philippines: Total CPI and Three Common Factors, 2001:Q1-2013:Q3

(In percent, year-on-year change)

Sources: IMF, World Economic Outlook; and IMF staff estimates.

Figure 15.Philippines: Actual Total Inflation and Common-Origin Inflation, 2001:Q1-2013:Q3

(In percent, year-on-year change)

Sources: IMF, World Economic Outlook; and IMF staff estimates.

Figure 16.Philippines; Actual Total Inflation and Idiosyncratic Inflation, 20G1:Q1-2G13:Q3

(In percent, year-on-year change)

Sources: IMF, World Economic Outlook, and IMF staff estimates.

Figure 17.Malaysia: Actual Inflation and Common-Origin Inflation, 2001:Q1-2013:Q3

(In percent, year-on-year change)

Sources: IMF, Wcrtd Economic Outlook: and IMF staff estimates.

Figure 18.Thailand: Actual Inflation and Common-Origin Inflation, 2001:Q1-2013:Q3

(In percent, year on year change)

Sources: IMF, World Economic Outlook; and IMF staff estimates.

Figure 19.Singapore: Actual Inflation and Common-Origin Inflation, 2001:Q1-2013:Q3

(In percent, year on year change)

Sources: IMF, World Economic Outlook; and IMF staff estimates.

Figure 20.Indonesia: Actual Inflation and Common-Origin Inflation, 2001:Q1-2013:Q3

(In percentyear-on-year change)

Sources: IMF, World Economic Outlook; and IMF staff estimates.

Figure 21.ASEAN-5: Common-Origin Inflation, 2001:Q1-2013:Q3

(In percent year-on-year change)

Sources: IMF, World Economic Outlook; and IMF staff estimates.

Idiosyncratic Inflation in the Philippines

20. From Figure 16, we find that idiosyncratic shocks have, on occasion, caused sizable deviations—both positive and negative—of common-origin inflation from total inflation in the Philippines. These country-specific factors were largest during the onset and immediate aftermath of the global financial crisis, when idiosyncratic factors initially pulled down inflation but later contributed to reflation. Thereafter, idiosyncratic inflation moderated and turned negative from late 2010-early 2013, but then turned positive.

D. Modeling Inflation in the Philippines: a Single-Country, Single Equation Approach

Phillips Curve Estimation

21. The common factors modeling of inflation suggests that inflation in the Philippines depends on world commodity price developments and movements in the U.S. dollar effective exchange rate. However, theory suggests that domestic cyclical conditions also matter. We assess the importance of these variables by estimating a Phillips curve augmented by world commodity prices and the nominal exchange rate for the period 2000–2013.5

22. The dependent variable is defined as the quarter-on-quarter percent change in headline CPI. The independent variables encompass: (i) lagged values of inflation to capture persistence; (ii) the one quarter lagged value of the output gap as a measure of cyclical conditions; and (iii) contemporaneous and lagged values of changes in global food and fuel prices (consistent with the large weight of these items in the CPI basket; and (iv) the nominal exchange rate. The impact of monetary and fiscal policies is therefore captured indirectly through their effect on the output gap. The inflation persistence term is included to capture explicit and de facto indexation of wages to inflation.

Model Specification

23. Global commodity prices are expressed in U.S. dollars and the bilateral exchange rate is denoted as number of pesos per U.S. dollar (i.e., an increase is a depreciation).6

Measurement of Contribution of Inflation Drivers

24. The cumulative contribution of the different inflation drivers can be expressed as:7

Estimation Results

Table 1.Philippines: Regression Analysis of Phillips Curve 1/2/
Dependent VariableCPI (In percent, period average, quarter over quarter)
Explanatory variables:
C0.0068

(0.0022) ***
CPI(−1)0.5577

(0.1556) ***
Global food0.0167

(0.0024) ***
Global fuel0.0134

(0.0023) ***
Exchange rate0.0315

(0.0313)
Output gap(−1)0.1878

(0.0550) ***
CPI(−2)−0.303

(0.1485) **
Global food (−1)0.0246

(0.0039) ***
Global fuel (−1)0.0173

(0.0042) ***
Exchange rate (−1)0.0309

(0.0184) *
CPI(−3)0.207

(0.0905) **
Global food (−2)−0.0219

(0.0064) ***
Global fuel (−2)0.0038

(0.0085)
Exchange rate (−2)0.0326

(0.0348)
CPI(−4)−0.1339

-0.1048
Global food (−3)0.0137

(0.0074) *
Global fuel (−3)−0.0039

(0.0025)
Exchange rate (−3)0.0366

(0.0209) *
Global food (−4)−0.0027

(0.0062)
Global fuel (−4)0.0005

(0.0043)
Exchange rate (−4)−0.003

(0.0209)
Sample:2000:Q1-2013:Q4Number of observations:56Adjusted R2:0.7578
Source: IMF staff estimates.

CPI: headline, period average, seasonally adjusted, log differenced quarter-over-quarter change; Output gap: in percent of potential real GDP, seasonally adjusted; Global food: proxied by global rice price index, period average, seasonally adjusted, differenced quarter-over-quarter change; Global fuel: proxied by global crude oil price index (simple average of three spot prices: Dated Brent, West Texas Intermediate, and the Dubai Fateh, 2005 = 100), period average, seasonally adjusted, log differenced quarter-over-quarter change; exchange rate: bilateral exchange rate, peso/US$, period average, log differenced quarter-over-quarter change.

Robust standard errors in parentheses.

indicates 1 percent,

5 percent,

10 percent statistical significance, respectively.

Source: IMF staff estimates.

CPI: headline, period average, seasonally adjusted, log differenced quarter-over-quarter change; Output gap: in percent of potential real GDP, seasonally adjusted; Global food: proxied by global rice price index, period average, seasonally adjusted, differenced quarter-over-quarter change; Global fuel: proxied by global crude oil price index (simple average of three spot prices: Dated Brent, West Texas Intermediate, and the Dubai Fateh, 2005 = 100), period average, seasonally adjusted, log differenced quarter-over-quarter change; exchange rate: bilateral exchange rate, peso/US$, period average, log differenced quarter-over-quarter change.

Robust standard errors in parentheses.

indicates 1 percent,

5 percent,

10 percent statistical significance, respectively.

25. Using the Akaike-Schwartz criterion, the optimal lag length is found to be four. Given the potential for serial correlation and heteroskedasticity, we use the Newey-West standard errors to find the consistent estimates.

  • The model fit is good and several of the variables within each category are significant and have the correct sign:

    • The model finds a high degree of autocorrelation in inflation, especially in the first quarter, with cumulative persistence of 33 percent over four consecutive quarters.8 The lagged output gap is also found to be significant, with a semi-elasticity with respect to inflation of 19 percent.

    • Global food price shocks are significant contemporaneously and in the first three lags. However, the cumulative pass-through to inflation is relatively small at 4.5 percent. Similarly, global fuel prices are significant contemporaneously and in the first lag, with a cumulative pass though of 4.6 percent.

    • While the exchange rate is not found to be significant contemporaneously, it has a significant and large impact in subsequent quarters. All told, the cumulative pass-through of the exchange rate to inflation is 19 percent.

Stability Analysis and Rolling Estimations

26. To assess the stability of the estimated model, we re-estimate using rolling 40-quarter windows, beginning in 1990:Q1. The dynamics of rolling coefficients (with ±2 NW standard errors) are displayed in Figure 22. The parameters are found to be quite unstable during the early time periods, but stabilize during the 2000s, coinciding with the implementation of the inflation targeting framework.

Figure 22.Estimated Coefficients, Rolling Sample with a Fixed Window of 40 Quarters

(First window: 1990:Q1-1999:Q4; last window: 2004:Q1-2013:Q4)

Source: IMF staff estimates.

Forecasting Inflation in the Philippines

27. In order to forecast future inflation, we first assess the model’s ability to forecast past inflation. Using rolling 40-quarter windows, we re-estimate the model and forecast inflation for the following 8 quarters, using the realized values for the explanatory variables. Figure 23 compares realized inflation with the model’s forecasts. While the model does not perform that well prior to 2000, predictive ability improves considerably thereafter.

Figure 23.Forecast Performance Assessment 1/: Rolling Samples, S-Quarter-Ahead Out-of-Sample Projection, 1991.Q4-2D13:Q4

(In percent, guarter-over-quarter change, period average)

Sources: Haver Analytics; and IMF staff estimates

1/ First estimation Window: 1980:Q1-1989:Q4; first forecast window:199:0Q1-1991:Q4(8-quarter-ahead out-of-sample);last estimation windows:2000:Q1-2011:Q4:first forecast window:2012:Q1-2013:Q4(8-quarter-ahead out-of-sample). Number of estimation windows:89:number of forecast windows:89.8-quarter-ahead rolling forecast series begins from 1991:Q4

28. Based on the parameters estimated over the period 2000:Q1-2013:Q4, we forecast future inflation during 2014:Q2 to 2019:Q4, using staff’s projections for the output gap and the nominal exchange rate, and WEO forecast for commodity prices (Figure 24). Inflation is expected to remain elevated in the near term, but to settle around 3.7 percent over the longer run, within the ±1 standard deviation interval spanning 2.7–4.7 percent. The long-run point estimate forecast lies above midpoint of the new inflation target band of 3±1 percent that will be introduced in 2015, with the upper range of the forecast lying above the band.

Figure 24.Philippines: Medium-Term Inflation Forecast 1/

(Forecast Horizon: 2014:Q2-2019:Q4)

Source IMF staff projections.

1/ 99 percent, 9j percent, 90 percent, and 67 percent confidence intervals are constructed using RMSE from rolling 8-quartert-ahead forecast exercise.

Appendix 1. Data in the Common Inflation Factor Model

We include 62 countries across the globe (listed below) in the dataset. The inflation rate therein refers to the year-on-year percent change of headline CPI, period average, at quarterly frequency. Data are seasonally adjusted and standardized.

List of Countries
Asia and

the Pacific
Latin America and the CaribbeanCEE and CISEuropeG-7Other Emerging

Market
DevelopingArgentinaRussiaEuro AreaUnited StatesIsrael
AsiaBrazilBelarus(excluding G-7)United KingdomSouth Africa
PhilippinesChileUkraineAustriaCanada
MalaysiaColombiaMoldovaBelgiumFrance
IndonesiaMexicoTurkeyCyprusGermany
ThailandPeruPolandEstoniaItaly
VietnamVenezuelaHungaryFinlandJapan
IndiaBulgariaGreece
ChinaRomaniaIreland
LithuaniaLatvia
NIEsSerbiaLuxembourg
SingaporeMalta
TaiwanNetherlands
Hong KongPortugal
KoreaSlovak
Slovenia
AdvancedSpain
Asia and Pacific
AustraliaEU Members
New Zealand(non-Euro Area)
Croatia
Czech
Denmark
Sweden
European AEs 2/
(excluding
EU members)
Iceland
Norway
Switzerland

CEE = Central and Eastern Europe; CIS = Commonwealth of Independent States.

AEs = Advanced economies.

CEE = Central and Eastern Europe; CIS = Commonwealth of Independent States.

AEs = Advanced economies.

References

    BaiJ. and S.Ng2013Principal Components Estimation and Identification of Static FactorsJournal of Econometrics Vol. 176 No. 1) pp. 1829.

    BaiJ. and S.Ng2002Determining the Number of Factors in Approximate Factor ModelsEconometrica Vol. 70 pp. 191221.

    BaiJ. and S.Ng2006Determining the Number of Factors in Approximate Factor Models, ErratamanuscriptColumbia University. Available via the Internet: http://www.columbia.edu/~sn2294/papers/correctionEcta2.pdf.

    BlanchardO. J. and J.Galí2007The Macroeconomic Effects of Oil Price Shocks: Why are the 2000s so Different from the 1970s?NBER Working Paper No. 13368 (Boston, Massachusetts: National Bureau of Economic Research).

    BussièreM. and T.Peltonen2008Exchange Rate Pass-Through in the Global Economy, The Role of Emerging Market EconomiesWorking Paper Series No. 951 (Frankfurt, Germany: European Central Bank).

    CiccarelliM. and B.Mojon2010Global InflationReview of Economics and Statistics Vol. 92 No. 3 pp. 524535.

    ChoueiriN.F.Ohnsorge and R.van Elkan2008Inflation Differentials in the EU: A Common (Factors) Approach with Implications for EU8 Euro Adoption ProspectsIMF Working Paper No. 08/21 (Washington: International Monetary Fund).

    De BandtO.A.Banerjee and T.Kozluk2007Measuring Long Run Exchange Rate Pass-ThroughEconomics Discussion Papers No. 2007–32 (Kiel, Germany: Kiel Institute for the World Economy). Available via the Internet: http://www.econorr.ics-ejournal.org/economics/discussionpapers/2007-32.

    HammondG.2012State of the Art of Inflation Targeting - 2012Centre for Central Banking Studies Handbook No. 29 (London: Bank of England). Available via the Internet: http://www.bankofengland.co.uk/education/Documents/ccbs/handbooks/pdf/ccbshb29.pdf.

    IhrigJ.M.Marazzi and A.Rothenberg2006Exchange-Rate Pass-Through in the G-7 CountriesInternational Finance Discussion Papers 851 (Washington: Board of Governors of the Federal Reserve System). Available via the Internet: http://www.federalreserve.gov/pubs/ifdp/2006/851/ifdp851.pdf.

    SargentT.J. and C.A.Sims1977Business Cycle Modeling Without Pretending to Have Too Much A-Priori Economic TheoryWorking Paper No. 55 (Minneapolis: Federal Reserve Bank of Minneapolis). Available via the Internet: www.mpls.frb.org/research/wp/wp55.pdf.

    StockJ.H. and M.W.Watson2002Forecasting Using Principal Components from a Large Number of PredictorsJournal of the American Statistical Association Vol. 97 No. 460 pp. 11671179.

    StockJ.H. and M.W.Watson1989New Indexes of Coincident and Leading Economic Indicators” in NBER Macroeconomics Annual 1989 Vol. 4 pp. 351409. Available via the Internet: http://www.nber.org/chapters/c10968.

Prepared by Huaizhu Xie.

For example, ordinary least squares (OLS) estimators are consistent only under the assumption that N/T (where N is the number of time series and T the number of observations) converges asymptotically to zero.

In technical terms, the number of common factors generated is equal to the number of time series; the common factors are the eigenvectors of equation (1) above; and the eigenvalue corresponding to each eigenvector indicates the marginal variance explained by that common factor.

Note that the loading for the United States itself is negative, consistent with U.S. dollar NEER appreciation contributing to lower prices for imported goods.

The estimation period is taken to begin in 2000, when the BSP adopted quasi inflation targeting, before it moved to formal inflation targeting in 2002. In this section, we ignore the impact of the great moderation of global inflation, which may be addressed through coefficient stability analysis.

We use the global rice price index from the IMF’s Global Primary Commodity Price database, given the large weight of rice in the CPI basket (nearly 9 percent, which represents nearly one quarter of food consumption).

These cumulative pass-throughs represent the long-run effect of a permanent change in oil and food price inflation when inflation has stabilized.

We discuss below that inflation persistence in the sample from 2000 onward is considerably smaller than during the 1990s.

Other Resources Citing This Publication